MoVerb

MoVerb数据集由明尼苏达大学计算机科学与工程系创建，专注于英语社交对话中的情态动词语义。该数据集包含4540条对话，每条对话由三个注释者使用Quirk和Palmer两种理论框架进行标注，总计27,240个情态动词语义注释。数据集的创建旨在解决情态动词在自然语言理解中的歧义问题，特别是在对话语境中的应用。通过比较不同理论框架下的注释一致性，本研究为情态动词的分类和理解提供了实证基础，有助于提升相关NLP任务的性能。

The MoVerb dataset was developed by the Department of Computer Science and Engineering at the University of Minnesota, focusing on the semantics of modal verbs in English social conversations. This dataset includes 4,540 dialogues, each annotated by three annotators using the two theoretical frameworks proposed by Quirk and Palmer, totaling 27,240 semantic annotations for modal verbs. The dataset was created to address the ambiguity of modal verbs in natural language understanding, particularly their application in conversational contexts. By comparing annotation consistency across different theoretical frameworks, this study provides an empirical basis for the classification and understanding of modal verbs, which helps enhance the performance of relevant NLP tasks.

搜集汇总

背景与挑战

背景概述

MoVerb数据集由明尼苏达大学创建，专注于英语社交对话中的情态动词语义，包含4540条对话和27,240个注释。该数据集通过Quirk和Palmer两种理论框架进行标注，旨在解决情态动词在自然语言理解中的歧义问题，并为相关NLP任务提供实证基础，提升性能。